摘要： 该文分析了一般主成分分析在处理非线性问题上存在的不足，阐述了基于核的主成分分析方法，并将其首次应用到综合评价中，应用结果表明：基于核的主成分分析方法具有优秀的特征提取性能，采用不同形式的核函数基本上能获得相同精度的特征提取效果。文末指出为提高运算速度、减少内存要求和降低复杂度，可在KPCA中结合主动学习的方法。

关键词: 特征提取, 核方法, 主成分分析, 综合评价

Abstract: This paper points out the drawbacks of the general PCA when it is used to solve nonlinear problem firstly. Then the kernel principal component analysis(KPCA) is introduced and used in the comprehensive evaluation. The research result shows that the KPCA has excellent performance of feature extraction and the KPCA with different kernel function can exhibit very similar accuracy. In the end of the paper, it indicate that active learning should be combined with KPCA for the purpose of improving the operation speed, reducing the demand of memory and decreasing the complexity.

中图分类号: 

• TP14